One could argue that the economic benefit per problem of the large computer would disappear as smaller computers improved. This would be true if smaller computers could be improved at a rate faster than the large computer. {...] Large computes are not at all limited in their rate of improvement, [...]
– J.F. Thornton, _Desing of a Computer: The Control Data 6600_, 1970
#ComputerArchitecture #ControlDataCorporation
That was absolutely true when it was written. It changed in the late 1980s, due to Moore's Law, and a little later, also due to the speed of light.